pep 484: explicit reexport as intended #4586
Conversation
See discussion on typing-sig. This doesn't take care of some third_party libraries, will follow up on those.
| @@ -268,7 +270,9 @@ class stat_result: | |||
| st_type: int | |||
|
|
|||
| if sys.version_info >= (3, 6): | |||
| from builtins import _PathLike as PathLike # See comment in builtins | |||
| from builtins import _PathLike | |||
|
|
|||
There was a problem hiding this comment.
I don't think this blank line helps readability.
| @@ -3,7 +3,9 @@ | |||
| import sys | |||
|
|
|||
| if sys.version_info < (3, 9): | |||
| from os import devnull as DEV_NULL | |||
| import os | |||
|
|
|||
third_party/2/six/__init__.pyi
Outdated
| @@ -5,7 +5,9 @@ import typing | |||
| import unittest | |||
| from __builtin__ import unichr as unichr | |||
| from functools import wraps as wraps | |||
| from StringIO import StringIO as BytesIO, StringIO as StringIO | |||
| from StringIO import StringIO as StringIO | |||
|
|
|||
|
Typeshed has surrendered itself to black, which mandates those blank lines |
|
Oooh, that last failing test is a fun one. The code flow through mypy is entirely different when defining a type alias instead of renaming on import. |
|
I haven't checked but I think you need to alias generics with something like |
|
I think this just works and needs builtins._PathLike to not be hardcoded in mypy. Going to attempt the merge dance. |
|
Yes; I was just about to say that I think it'd be weird to have to add typevars to emulate behavior that would have previously worked in the " |
As a result of python/typeshed#4586
As a result of python/typeshed#4586
…9512) As a result of python/typeshed#4586 Co-authored-by: hauntsaninja <>
…ython#9512) As a result of python/typeshed#4586 Co-authored-by: hauntsaninja <>
See discussion on typing-sig.
This doesn't take care of some third_party libraries, will follow up on
those.